Re: st: score test for the parallel regression assumption

[brendan.halpin@ul.ie (Brendan Halpin)]

On Tue, May 22 2012, Thomas VanHeuvelen wrote:

> Does anyone know of a way to perform a score test for the parallel
> regression assumption after running an ordered logistic regression command
> (e.g. ologit or gologit2)? More specifically, is anyone aware of a
> program/procedure in Stata that can reproduce the score test given by
> SAS's -proc logistic- command?

I don't know what the score test is, but does the Brant test do what
you want? It's part of Long & Freese's -spostado- (-findit spostado- to
get the details).

Example:

. xi: ologit ropfamr i.rsex rage
i.rsex            _Irsex_1-2          (naturally coded; _Irsex_1 omitted)

Iteration 0:   log likelihood = -18924.158  
Iteration 1:   log likelihood = -17818.231  
Iteration 2:   log likelihood = -17802.121  
Iteration 3:   log likelihood = -17802.088  
Iteration 4:   log likelihood = -17802.088  

Ordered logistic regression                       Number of obs   =      12725
                                                  LR chi2(2)      =    2244.14
                                                  Prob > chi2     =     0.0000
Log likelihood = -17802.088                       Pseudo R2       =     0.0593

------------------------------------------------------------------------------
     ropfamr |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    _Irsex_2 |   .8339045    .033062    25.22   0.000     .7691041    .8987048
        rage |  -.0371618   .0009172   -40.51   0.000    -.0389595    -.035364
-------------+----------------------------------------------------------------
       /cut1 |  -3.833869   .0597563                     -3.950989   -3.716749
       /cut2 |  -2.913506   .0547271                      -3.02077   -2.806243
       /cut3 |  -1.132863   .0488522                     -1.228612   -1.037115
       /cut4 |   .3371151   .0482232                      .2425994    .4316307
------------------------------------------------------------------------------

. brant, detail

Estimated coefficients from j-1 binary regressions

                 y>1         y>2         y>3         y>4
_Irsex_2   1.0198492   .91316651   .76176797    .8150246
    rage  -.02716537  -.03064454  -.03652048  -.04571137
   _cons   3.2067856   2.5225826   1.1214759  -.00985108

Brant Test of Parallel Regression Assumption

    Variable |      chi2   p>chi2    df
-------------+--------------------------
         All |    101.13    0.000     6
-------------+--------------------------
    _Irsex_2 |     15.88    0.001     3
        rage |     81.07    0.000     3
----------------------------------------

A significant test statistic provides evidence that the parallel
regression assumption has been violated.



Regards,

Brendan 
-- 
Brendan Halpin,   Department of Sociology,   University of Limerick,   Ireland
Tel: w +353-61-213147  f +353-61-202569  h +353-61-338562;  Room F1-009 x 3147
mailto:brendan.halpin@ul.ie    ULSociology on Facebook: http://on.fb.me/fjIK9t
http://teaching.sociology.ul.ie/bhalpin/wordpress         twitter:@ULSociology
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/